1 |
On 11/14/11 09:25, Alex Alexander wrote: |
2 |
> On Sun, Nov 13, 2011 at 05:59:21PM -0800, Zac Medico wrote: |
3 |
>> Well, it's much easier to gather interest and get feedback if we deploy |
4 |
>> the change and ask questions later. |
5 |
> What if we tried solving this problem by providing more options instead |
6 |
> of trying to guess what the users want? :) |
7 |
> |
8 |
> Imagine the following output (when jobs == 1): |
9 |
> |
10 |
>>>> Verifying ebuild manifests |
11 |
>>>> Emerging (1 of 1) www-client/chromium-16.0.912.36 |
12 |
>>>> Quiet building enabled. Enable for [P]ackage or [S]ession. [L]earn more. |
13 |
>>>> Jobs: 0 of 1 complete, 1 running Load avg: 0.23, 0.18, 0.10 |
14 |
> Pressing P would only show the log for the actively built package. |
15 |
> Pressing S would show all the logs for this session, starting with the |
16 |
> active one. |
17 |
> |
18 |
> Pressing L would print out a short set of instructions, something |
19 |
> like: |
20 |
> |
21 |
> "To make portage output easier to track and understand, --quiet-build |
22 |
> has been enabled by default. You may restore the old, verbose behavior |
23 |
> temporarily by using the P and S commands, or permanently by adding |
24 |
> '--quiet-build=n' to your make.conf's EMERGE_DEFAULT_OPTS." |
25 |
> |
26 |
> I believe many users would appreciate the ability to output logs on |
27 |
> demand. :) |
28 |
Interactive stuff is a bad idea. Please don't make me figure out ways to |
29 |
interactively stab you ;) |
30 |
(I already spent so much time getting interactive ebuilds into a |
31 |
manageable state, and that was a more futile discussion than this one) |
32 |
|
33 |
Why do y'all want to make it harder for me to figure out |
34 |
|
35 |
1) is it just slow at compiling or stuck in a loop? |
36 |
2) did it compile with proper settings ( just because the ebuild |
37 |
finished with exit code 0 doesn't mean it's correct - yesterday I had to |
38 |
turn off quiet-build and re-compile squashfs-tools just to see if the |
39 |
build output was as expected) |
40 |
3) did it install the files I expected it to install? (hmm, why don't I |
41 |
have the "znurgh" binary now?!) |
42 |
|
43 |
Not all failures are detected build failures ... |
44 |
|
45 |
And handling it over EMERGE_DEFAULT_OPTS is also ugly, if I had anything |
46 |
else in there I'd find it quite annoying to figure out how to keep --ask |
47 |
--keep-going while disabling --quiet-build selectively. |
48 |
So it'd be nicer to have a PORTAGE_QUIET_BUILD or similar var in |
49 |
make.conf - and still I'm annoyed that I have to spend time unbreaking |
50 |
the defaults. |
51 |
|
52 |
Can we please get this reset to "classic mode" and not try to innovate |
53 |
things that worked perfectly well the last decade? |
54 |
|
55 |
|
56 |
Patrick |